Lectures on FWI 3: Theoretical Background II

Welcome back to my short series of "Lectures on Full Waveform Inversion". In this lecture, I will improve the convergence performance of the FWI algorithm, introduced in the last lecture, by replacing the steepest descent with the conjugate gradient and quasi-Newton l-bfgs method. I will also discuss the drawbacks of using local optimization problems in case of multi-modal objective functions. The short series "Lectures on FWI" is part of the "Inverse Theory" course in the masters program "Geophysics" at Christian-Albrechts-University Kiel.